Rapport de mission useR!2017 - Bruxelles

1.2.1 Tidy way of thinking (#dplyr #tidyr #ggplot @tom @pat)

1.2.2 Reproductible science (#rmarkdown)

1.2.3 Cartographie (#meteo #satellite #mapping #openstreetmap #dynamicmap)

1. Cours pour le traitement de données spatiale (ULg, Lisboa, Montpellier)
   

   http://www.openspat.eu/

2. open data météo Allemagne : (Deutsh Weather Data)
   
   • a package R for that : rdwd
   • leaflet
   • https://github.com/brry/prectemp
3. New direction : sf package
   

   Important [à creuser!, see details Spatial Data in R: New Directions ]

   Avis sur les outils :

   • "sf is the future" (en remplacement de sp) : plus de caractères à écrire, mais plus flexible et plus précis .
   • pour les cartes statiques : "ggplot2 is the future", préférez geom_polygon à geom_map. (ggviz : avis négatif).
   • pour les cartes dynamiques : leaflet, avis ++, activement maintenu par RStudio (shiny : avis pas enthousiaste pour les cartes).
4. Package lié à OSM, … - Maps are data, so why plot data on a map?
   

   osmplotr, osmdata (see details)

   Présentation EXTRA !!: https://schd.ws/hosted_files/user2017/88/padgham.pdf

   • vidéo : https://channel9.msdn.com/Events/useR-international-R-User-conferences/useR-International-R-User-2017-Conference/Maps-are-data-so-why-plot-data-on-a-map

   à suivre : premier test ok via le script R file://FILESERVER/Commun/Documentation/Statistiques/ScriptExemples/essaicartoosmplotr.R (dossier réseau)

   Mais pas encore beaucoup de fonctionnalités (??), test plus avancé non concluant. ) creuser.

5. linking R and other GIS (QGIS/GRASS/…)
   
   • Package RQGIS : http://sched.co/AxpV
   • Package Link2GI : http://sched.co/Axpb
6. Raster et satellite : développement de STARS
   

   Scalable, Spatiotemporal Tidy Arrays for R (stars)

   http://sched.co/AxpE

1.2.4 Autres thèmes importants (mais non creusés, à creuser)

1. Parallel computing
   
2. Modules in R
   

   voir détails (dernier jour), c'est entre un script et un package

1.3.1 KEYNOTE: Teaching data science to new useRs

1.3.2 A Spreadsheet for R : Jamovi (@Mary, @Patoche, …)

1.3.3 URBAN GREEN SPACES AND THEIR BIOPHONIC SOUNDSCAPE COMPONENT

1.3.4 Dose-response (@Gilles)

1.3.5 text mining

1.3.6 Morphological Analysis in R

1.4.1 Programme et liens vers les ressources

1.4.2 Vidéos des présentations, slides + son :

1.4.3 Autres liens

2.1.1 Geospatial Data Visualization Using R

2.1.2 Spatial Data in R: New Directions

1. In comparison to sp, package sf
   
   • implements all types and classes of simple features
   • has no support for gridded (raster) data
   • uses data.frames for features, and list columns for feature geometry
   • uses S3 instead of S4
   • is also built on top of GDAL, GEOS and Proj.4
   • tries to provide a simpler and cleaner API (or user experience) conversions to/from sp makes it still easy to work “backwards”
2. What’s really new, and better in sf, compared to sp?
   
   • simple features is a widely adopted standard
   • tidyverse compatibility
   • ggplot2 support (install_github, under development)
   • support for measurement units
   • partial support for computations using geographic coordinates
   • support by mapview, tmap, mapedit; 15 revdeps on CRAN
   • binary geom ops: fast (indexed), low memory footprint; flexible st_join
   • (c)lean Rcpp interface to external dependencies GDAL/GEOS/Proj.4
   • fast WKB (de)serialization, in C++

2.1.3 Data Carpentry – Open and Reproducible Science with R

1. Four facets of reproducibility
   

   Documentation

   explanation and commenting of why and how an analysis is carried out in human readable language

   Organization

   tools to organize your projects so that you don't have a single folder with hundreds of files

   Automation

   the power of scripting to create automated (and self documenting) data analyses

   Dissemination

   publishing is not the end of your analysis, rather it is a way station towards your future research and the future research of others

2. Three key principles for (file) names
   
   1. Machine readable
      • easy to search for files later
      • easy to narrow file lists based on names
      • easy to extract info from file names (regexp-friendly)
   2. Human readable
      • name contains information on content, or
      • name contains semantics (e.g., place in workflow)
   3. Plays well with default ordering
      • use numeric prefix to induce logic order
      • left pad numbers with zeros
      • use ISO 8601 standard (YYYY-mm-dd) for dates
3. Organisation of the R folder
   

   Some organizing principles for files

   data-raw

   the original data, do not edit or directly alter any of the files in this folder.

   data-output

   intermediate datasets that will be generated by the analysis. We write them to CSV or other portable formats, particularly useful when it take a long time (or expensive resources) to generate.

   fig

   where we can store the figures used in the manuscript.

   R

   our R code (the functions)

   • For large projects it often becomes easier to keep the prose separated from the code.
   • If you have a lot of code (and/or manuscript is long), it's easier to navigate.
   • Or just create an R package.

   tests

   the code to test that our functions are behaving properly and that all our data is included in the analysis.

   |
   +-- data-raw/
   |   |
   |   +-- gapminder-5060.csv
   |   +-- gapminder-7080.csv.csv
   |   +-- ....
   |
   +-- data-output/
   |
   +-- fig/
   |
   +-- R/
   |   |
   |   +-- figures.R
   |   +-- data.R
   |   +-- utils.R
   |   +-- dependencies.R
   |
   +-- tests/
   |
   +-- manuscript.Rmd
   +-- make.R
   

4. Licences
   

   http://journals.plos.org/ploscompbiol/article/file?id=10.1371/journal.pcbi.1002598&type=printable

2.1.4 Efficient R programming

2.2.1 Sponsor talk : Microsoft

1. History :
   
   • We never heard about R (5th language)
   • R does not do what I need (number of packages !)
   • We can't use R because it is opensource (Microsoft !)
   • We can't use R in a regulated environment (FDA !)
   • R is too slow (Microsoft server, SQL)
   • R is not deeling with big data (spark, …)
   • You can't use R in production : trains, planes, planes, chemicals, …
   • We can't get support for R, Who stands behind R? (microsoft, …)

   now R ecosystem

2.2.2 Keynote : Structural Equation Modeling

2.2.3 Shiny app for 700 end users

2.2.4 sonification of data

2.2.5 sports data

2.2.6 soundecology

2.2.7 Maps are data, so why plot data on a map?

2.2.8 Text mining session

2.2.9 R and GIS session

1. Jannes Muenchow
   

   qgis and R : interaction dans r : ce qui manque : des geoalgorithme

   RSAGA RQGIS RGRASS …

   Qgis : python API lots of geoalgorithm and python console then also R via reticulate (apen_app())

   reticulate = interface of python on R

   use functions : find_algorithms(), get_usage()

   library(RGIS) Jannes Muenchow (2013) : plant diversity prediction

2. Ege Rubak : package S2
   

   Interfacing Google's spherical geometry library (S2) for spatial data in R

   It both facilitates geometric operations directly on the sphere such as polygonal unions, intersections, differences etc. without the hassle of projecting data in the common latitude and longitude format, and provides an efficient quadtree type hierarchical geospatial index.

3. Scalable, Spatiotemporal Tidy Arrays for R (stars)
   

   Edzer Pebesma anticipate the queston : raster include simple features ask collaboration for developping the package

2.3.1 Sponsors Talk : OpenAnalytics

2.3.2 Plenary : dose-reponse

2.3.3 Networks analysis (NL)

2.3.4 JAMOVI

2.3.5 data journalism

2.3.6 Lightning talks - ecology (really small talk !)

1. deutch weather data
   

   a package R for that : rdwd leaflet github.com/brry/prectemp

2. Earth movement
   

   package ESEIS toolbox to weld geomorphic, seismologic, spatial and time series analysis

   www.micha-dietze.de

3. WEF : water-energy-food
   

   connection, real important for the future swat model, best management practices

4. subsurface hydrology
   

   critic : non reproducable, than is it science ?

5. PM2.5 in India
   
6. Andy South (@southmapr)
   

   rnaturalearth sustainability, (soft. sust. institute) plot(ne_countries(country="belgium",scale="medium")

7. open spat
   

   Ulg, lisboa, Montpellier supagro très intéressant !!! +

   Learning modules to get skills on free tools

   analyze and interpret spatial data

   prequel online course

   Rmarkdown

   • and scenari (free !)(check !!)

   http://openspat.eu/

8. Tobias Gauster - package smires
   

   European project on Science and Management of Intermittent Rivers and Ephemeral Streams

2.4.1 KEYNOTE: R tools for the analysis of complex heterogeneous data

2.4.2 tidyverse grammar Tidy evaluation

2.4.3 modules

2.4.4 R and Haskell: Combining the best of two worlds